Photogrammetric Engineering & Remote Sensing · 1989 · 10 citations · 0 references
ComputingCivil EngineeringDigital Elevation ModelingFinite Element AnalysisTerrain AnalysisDigital Elevation ModelsDigital Signal ProcessingDigital Terrain ModelingDigital Terrain ModelsTerrain ModelingTerrain Surface ModelingDigital Terrain Model GenerationDigital Terrain RepresentationSpeedTerrain ReconstructionTerrain Roughness AnalysisTerrain Roughness ModelingTerrain Surface Estimation
The development and results of some array relaxation algorithms of fast array algebra solutions for automated validation and progressive sampling of digital terrain models (DTM) are described. Array algebra DTM techniques are reviewed, including new formulations of finite element DTM of large systems of linear and non-linear equations. The array relaxation algorithm is introduced as a generalized fast transform and signal processing technique for a computational solution of general non-separable and shift variant systems. The solution speed of the algorithm is thousands of nodes per second in inexpensive general purpose computers such as microprocessors and low-end work stations. The high DTM solution speed of array relaxation opens up the practical feasibility of on-line sensing of blunders, determination of local terrain roughness, and updated modeling accuracy of progressive sampling. The reliability of conventional single point two-ray stereo correlation is thereby improved, and automation of most of the slow and expensive manual edit process greatly improves overall production speed, quality, and economy of DTM generation. Some results of practical DTM experiments are reported in this paper to prove the practical feasibility of automated on-line and off-line validation and progressive sampling of the array relaxation algorithm. The generalization of the array relaxation algorithm to the new array correlation technique of multi-ray global least-squares correlation is also discussed. This efficient and automated multi-image mensuration technique is a part of a new photogrammetric systems concept, positioning, of the array algebra technology. The compiler positioning system is expanding the philosophical and computational foundations of photogrammetry and related surveying sciences to the automated systems of array algebra digital photogrammetry with a profound impact on the speed, quality, and economy of future mapping systems.